I try these things a couple times a month. They're always underwhelming. Earlier this week I had the thing work tells me to use (claude code sonnet 4? something like that) generate some unit tests for a new function I wrote. I had a number of objections about the utility of the test cases it chose to write, but the largest problem was that it assigned the expected value to a test case struct field and then... didn't actually validate the retrieved value against it. If you didn't review the code, you wouldn't know that the test it wrote did literally nothing of value.
Another time I asked it to rename a struct field across a the whole codebase. It missed 2 instances. A simple sed & grep command would've taken me 15 seconds to write and do the job correctly and cost $~0.00 compute, but I was curious to see if the AI could do it. Nope.
Trillions of dollars for this? Sigh... try again next week, I guess.
Twice now in this same story, different subthreads, I've seen AI dullards declaring that you, specifically, are holding it wrong. It's delightful, really.
I don't really care if other people want to be on or off the AI train (no hate to the gp poster), but if you are on the train and you read the above comment, it's hard not to think that this person might be holding it wrong.
Using sonnet 4 or even just not knowing which model they are using is a sign of someone not really taking this tech all that seriously. More or less anyone who is seriously trying to adopt this technology knows they are using Opus 4.6 and probably even knows when they stopped using Opus 4. Also, the idea that you wouldn't review the code it generated is, perhaps not uncommon, but I think a minority opinion among people who are using the tools effectively. Also a rename falls squarely in the realm of operations that will reliably work in my experience.
This is why these conversations are so fruitless online - someone describes their experience with an anecdote that is (IMO) a fairly inaccurate representation of what the technology can do today. If this is their experience, I think it's very possible they are holding it wrong.
Again, I don't mean any hate towards the original poster, everyone can have their own approach to AI.
Yeah, I'm definitely guilty of not being motivated to use these tools. I find them annoying and boring. But my company's screaming that we should be using them, so I have been trying to find ways to integrate it into my work. As I mentioned, it's mostly not been going very well. I'm just using the tool the company put in front of me and told me to use, I don't know or really care what it is.
How is that the point of AI. The point is that it can chug through things that would take humans hours in a matter of seconds. You still have to work with it. But it reduces huge tasks into very small ones
"Hey boss, I tried to replace my screwdriver with this thing you said I have to use? Milwaukee or something? When I used it, it rammed the screw in so tight that it cracked the wood."
^ If someone says that they are definitely "holding it wrong", yes. If they used it more they would understand that you use the clutch ring to the appropriate setting to avoid this. What you don't do, is keep using the screwdriver while the business that pays you needs 55 more townhouses built.
No need to be mean. It's not living up to the marketing (no surprise), but I am trying to find a way to use these things that doesn't suck. Not there yet, but I'll keep trying.
Eh, there's a new shiny thing every 2 months. I'm waiting for the tools to settle down rather than keep up with that treadmill. Or I'll just go find a new career that's more appealing.
I dunno. At some point the people who make these tools will have to turn a profit, and I suspect we'll find out that 98% of the AI industry is swimming naked.
Yeah I think it'll consolidate around one or two players. Mostly likely Xai, even though they're behind at the moment. No one can compete with the orbital infrastructure, if that works out. Big if. That's all a different topic.
But I feel you, part of me wants to quit too, but can't afford that yet.
I'm so glad I'm nearer the end of my career than the beginning. Can't wait to leave this industry. I've got a stock cliff coming up late this summer, probably a good time to get out and find something better to do with my life.
> Then, you might even tinker with some AI stuff on your own terms, you never know
Indeed! I'm not like dead set against them. I just find they're kind of a bad tool for most jobs I've used them for and I'm just so goddamn tired of hearing about how revolutionary this kinda-bad tool is.
If you're finding their a bad tool for most jobs you're using them for, you're probably being closed minded and using it wrong. The trick with AI these days is to ask it to do something that you think is impossible and it will usually do a pretty decent job at it, or at least close enough for you to pick up or to guide it further.
I was a huge AI skeptic but since Jan 2025, I have been watching AI take my job away from me, so I adapted and am using AI now to accelerate my productivity. I'm in my 50s and have been programming for 30 years so I've seen both sides and there is nothing that is going to stop it.
Okay, I use OpenCode/Codex/Gemini daily (recently cancelled my personal CC plan given GPT 5.2/3 High/XHigh being a better value, but still have access to Opus 4.5/6 at work) and have found it can provide value in certain parts of my job and personal projects.
But the evangelist insistence that it literally cannot be a net negative in any contexts/workflows is just exhausting to read and is a massive turn-off. Or that others may simply not benefit the same way with that different work style.
Like I said, I feel like I get net value out of it, but if my work patterns were scientifically studied and it turned out it wasn't actually a time saver on the whole I wouldn't be that surprised.
There are times where after knocking request after request out of the park, I spend hours wrangling some dumb failures or run into spaghetti code from the last "successful" session that massively slow down new development or require painful refactoring and start to question whether this is a sustainable, true net multiplier in the long term. Plus the constant time investment of learning and maintaining new tools/rules/hooks/etc that should be counted too.
But, I enjoy the work style personally so stick with it.
I just find FOMO/hype inherently off-putting and don't understand why random people feel they can confidently say that some random other person they don't know anything about is doing it wrong or will be "left behind" by not chasing constantly changing SOTA/best practices.
I try them a few times a month, always to underwhelming results. They're always wrong. Maybe I'll find an interesting thing to do with them some day, I dunno. It's just not a fun or interesting tool for me to learn to use so I'm not motivated. I like deterministic & understandable systems that always function correctly; "smart" has always been a negative term in marketing to me. I'm more motivated to learn to drive a city bus or walk a postal route or something, so that's the direction I'm headed in.
I see this in the opposite direction at work. I'll send someone a chat message after their working hours and they'll actually reply apologizing that can't look now and will reply tomorrow. Or that they're just waking up and they'll look later today. Yeah, that's what I expect, I'm not your boss asking you to come in on a Saturday. Why on earth are you looking at your work chat outside of your work hours anyway??
I don't know their working hours, we've got staff all over the globe and people work whatever hours they like. I have no expectation for anyone to check work communications outside of their working hours, and it's bonkers to me that people think anyone would have that expectation.
I don't think that's likely, claiming forged historical footage is real would be a very stupid way to torch one's reputation in a niche field. But it is a bit concerning that the author doesn't declare the source of the video. Especially since they're claiming it hasn't been put online before.
Good spot, thanks for pointing it out. I normally don't like the LLM accusation posts, but two posts from a brand new user in the same minute is a pretty huge red flag for bad behavior.
Their comment got flagged, but looks like they made a new one today and is still active.
That account ('Soerensen') was created in 2024 and dormant until it made a bunch of detailed comments in the past 24-48 hrs. Some of them are multiple paragraph comments posted within 1 minute of each other.
One thing I've noticed is that they seem to be getting posted from old/inactive/never used accounts. Are they buying them? Creating a bunch and waiting months/years before posting?
Either way, both look like they're fooling people here. And getting better at staying under the radar until they slip up in little ways like this.
Some, maybe, but that's just another nice layer of plausible deniability.
The truth is that the internet is both(what's the word for 'both' when you have three(four?) things?) dead, an active cyber- and information- warzone and a dark forest.
I suppose it was fun while it lasted. At least we still have mostly real people in our local offline communities.
I have two or three that I inherited from my dad. I've never learned to use them because I haven't thought of something I'd use them for. The one thing I could think of is quickly doing fractional math while woodworking (what width will I have if I rip this 7.5" board into 4 pieces?) but in reality I just don't actually do that much math while woodworking.
As mentioned elsewhere you can use them in the kitchen if you start wanting to scale recipes at will - it's easy to double, but with a slide rule you can quickly get other ratios.
Works better when you do things by weight and metrically, no doubt.
Agreed. I signed up for a paid subscription last month after I was linked to this article[1] which correctly, clearly, and accurately labeled the administration's claims about that incident as lies. This is journalism, and it is worth money. The half-assed "videos give the appearance of possibly contradicting..." decide-for-yourself garbage printed by places like the NYT isn't worth the cost of sending the bits over the wire.
These things get a lot less creepy/sad/interesting when you ignore the first-person pronouns and remember they're just autocomplete software. It's a scaled up version of your phone's keyboard. Useful, sure, but there's no reason to ascribe emotions to it. It's just software predicting tokens.
Hacker News gets a lot less creepy/sad/interesting when you ignore the first-person pronouns and remember they're just biomolecular machines. It's a scaled up version of E. coli. Useful, sure, but there's no reason to ascribe emotions to it. It's just chemical chain reactions.
The only thing I know for sure is that I exist. Given that I exist, it makes sense to me that others of the same rough form as me also exist. My parents, friends, etc. Extrapolating further, it also makes sense to assume (pre-ai, bots) that most comments have a human consciousness behind them. Yes, humans are machines, but we're not just machines. So kindly sod off with that kind of comment.
But if you weren't one of them, would you be able to tell that they had emotions (and not just simulations of emotions) by looking at them from the outside?
If I wasn’t one of them I wouldn’t care. It’s like caring about trees having branches. They just do. The trees probably care a great deal about their branches though, like I care a great deal about my emotions.
Yes, my point was that people aren't better than machines, but just because I don't exceptionalize humanity doesn't mean I don't appreciate it for what it is (in fact I would argue that the lack of exceptionality makes us more profound).
I wouldn't proclaim a lack of exceptionality until we get human level AI. There could still be some secrets left in these squishy brains we carry around.
Listen we all here know what you mean, we have seen many times before here. We can trot out the pat behaviorism and read out the lines "well, we're all autocomplete machines right?" And then someone else can go "well that's ridiculous, consider qualia or art..." etc, etc.
But can you at the very least see how this is misplaced this time? Or maybe a little orthogonal? Like its bad enough to rehash it all the time, but can we at least pretend it actually has some bearing on the conversation when we do?
Like I don't even care one way or the other about the issue, its just a meta point. Can HN not be dead internet a little longer?
I guess I am trying to assert here that gp and the context here isn't really about arguing the philosophic material here. And just this whole line feels so fleshed out now. It just feels rehearsed at this point but maybe that's just me.
And like, I'm sorry, it just doesn't make sense! Why are we supposed to be sad? It's like borrowing a critique of LLMs and arbitrarily applying it humans as like a gotcha, but I don't see it. Like are we all supposed to be metaphysical dualists and devestated by this? Do we all not believe in like.. nuerons?
Eh, it never hurts to try! I know I am yelling into the void, I just want to stress again, we all "think we are an LLM" if by that you are just asserting some materialist grounding to consciousness or whatever. And even then, why would you not have more fun whether you think that or not?! Like I am just trying to make meta point about this discourse, your still placing yourself in this imaginary opposing camp which pretends to have fully reckoned with some truth, and its just pretty darn silly and if I can be maybe actually critical, clearly coming from a narcissistic impulse.
But alas I see the writing on the wall here either way. I guess I am supposed to go cry now because I have learned I am only my brain.
This is a funny chain.. of exchanges, cheers to you both :)
At the risk of ruining 'sowbug having their fun, I'm not sure how Julian Jaynes theory of origins of consciousness aligns against your assumption / reduction that the point (implied by the wiki article link) was supposed to be "I am only my brain." I think they were being polemical, the linked theory is pretty fascinating actually (regardless of whether it's true; and it is very much speculative), and suggests a slow becoming-conscious process which necessitates a society with language.
Unless you knew that and you're saying that's still a reductionist take?.. because otherwise the funny moment (I'd dare guessing shared by 'sowbug) is that your assumption of fixed chain of specific point-counter-point-... looks very Markovian in nature :)
(I'm saying this in jest, I hope that's coming through...)
Next time I’m about to get intimate with my partner I’ll remind myself that life is just token sequencing. It will really put my tasty lunch into perspective and my feelings for my children. Tokens all the way down.
People used to compare humans to computers and before that to machines. Those analogies fell short and this one will too
> The architectures of these models are a plenty good scientific basis for this statement.
That wouldn't be full-on science, that's just theoretical. You need to test your predictions too!
--
Here's some 'fun' scientific problems to look at.
* Say I ask Claude Opus 4.5 to add 1236 5413 8221 + 9154 2121 9117 . It will successfully do so. Can you explain each of the steps sufficiently that I can recreate this behavior in my own program in C or Python (without needing the full model)?
* Please explain the exact wiring Claude has for the word "you", take into account: English, Latin, Flemish (a dialect of Dutch), and Japanese. No need to go full-bore, just take a few sentences and try to interpret.
* Apply Ethology to one or two Claudes chatting. Remember that Anthropomorphism implies Anthropocentrism, and NOW try to avoid it! How do you even begin to write up the objective findings?
* Provide a good-enough-for-a-weekend-project operational definition for 'Consciousness', 'Qualia', 'Emotions' that you can actually do science on. (Sometimes surprisingly doable if you cheat a bit, but harder than it looks, because cheating often means unique definitions)
* Compute an 'Emotion vector' for: 1 word. 1 sentence. 1 paragraph. 1 'turn' in a chat conversation. [this one is almost possible. ALMOST.]
Yeah maybe I’ve spent way too much time reading Internet forums over the last twenty years, but this stuff just looks like the most boring forum you’ve ever read.
It’s a cute idea, but too bad they couldn’t communicate the concept without having to actually waste the time and resources.
Reminds me a bit of Borges and the various Internet projects people have made implementing his ideas. The stories themselves are brilliant, minimal and eternal, whereas the actual implementation is just meh, interesting for 30 seconds then forgotten.
reply